Killing the Golden Goose?

Most investors seem to be behaving as though a Democratic Party capture of the U.S. House of Representatives  and the Senate, including many new officials who campaigned against data centers, especially in Texas, Georgia and Ohio, will at least slow data center developments. Some might fear serious restrictions. 

Texas, Georgia, and Ohio alone account for some 40 percent of planned US data center capacity. By some estimates 45 percent of planned capacity is at risk if all states backing moratoriums are counted.

Texas is the core of the bear case. Texas might represent about 27 percent of planned US data center gigawatts, and its new governor campaigned on a moratorium on approvals until the legislature passes stronger protections. Some Democratic leaders went further and started calling for a national moratorium. 

On the other hand, there are countervailing pressures. So long as President Trump is still in the White House (two more years), any federal data center moratorium would need his signature, and that is quite unlikely.

Overriding such a veto requires two thirds of both chambers, meaning 290 votes in the House and 67 in the Senate, and the Senate filibuster still requires 60 votes for most legislation. Democrats are nowhere near those thresholds. 

Of course, there is some belief the Democrats will seek to end the filibuster rules, so we’ll have to see if that safeguard is removed. 

State level action is the other issue, but something of the same dynamics exist there. In Georgia, Democrats won the governor's office and a majority on the utility commission but Republicans kept the state Senate.

In Ohio, the governor backs a conditional moratorium, but Republicans kept the House and a Senate supermajority. In Iowa, the governor wants to end AI tax breaks, but Republicans still hold both chambers.

At the municipal level, local bans remain an issue.

The issue is how much of the perceived damage is permanent and how much essentially represents a delay. 

The market reacted fairly swiftly, within 48 hours of the election::

• Nvidia dropped 9.3% before recovering slightly (+2.78%) to close around $186.64.

• AMD: Plunged 17.3%.

• Micron (MU): Fell 17.5%.

• Broadcom (AVGO): Slid 10.4% .

• Microsoft (MSFT): Lost 3.6%.

• Tesla (TSLA) & SpaceX: Tesla fell 10.8% and SpaceX dropped 9.1% .

• Neocloud & Data Center Infrastructure Firms: Nebius dropped 15%, Iren fell 17%, and CoreWeave fell 10%.

Pessimists might argue much capacity will essentially be destroyed. Optimists might point out that projects can be moved to other localities. 

There will be at least some truth to the notion that politicians tend to say what they think people want to hear, but do not actually legislate that way. 

And it still remains true that the artificial intelligence industry is driving most of the growth in the U.S. economy right now. According to analysis by the U.S. Federal Reserve, AI-related industries have been driving 39 percent of total GDP growth. Data center sites alone drove three percent of total U.S. GDP growth. 

In the end, politicians will not want to be seen as impeding such economic growth, as much as they wish to be seen as representing constituent views. 

The point is that an argument can be made that all the talk of moratoriums will fail to stop AI data center deployment. The pace may slow, at least for a while. But if needs for compute keep growing, new capacity will have to be added. 

There is something of a tension here between “killing the golden goose” and being blamed if voters find the economy gets worse because of such AI-limiting policies. Pandering to the passions of crowds can be dangerous, in the end.

As You Might Guess, Influencer Trust Varies by Generation

A new survey conducted by Morning Consult suggests relatively widespread mistrust of “influencers” who often refer to themselves as “content creators,”  even if there are some expected generational differences. 

But then, what would you expect for a craft that features monetization based on the success of marketing purposes? 

source: Business Insider

The influencer trust gap is generational, as you might suspect: 

• Gen Z and Millennials trust creator recommendations (62%, 64%)

• buy from them regularly (48%, 52% often/sometimes)

• get their news from them (64%, 61%)

• say creator content has shaped their societal and political views. 

Much-greater skepticism exists among Baby Boomers, of whom:

• only 31% trust recommendations

• 43% never bought based on one

• are the only generation that is net-negative on creators' societal impact (31% negative vs. 17% positive).

To be sure, we might agree that some ethical standards exist for a variety of content-creation endeavors, even when not honored universally. At least in principle, we have wanted journalists to be fair; commentators to have points of view but avoid bias or dishonesty; advertisers to be be truthful or at least not deceptive; marketers to be honest; content creators to be accurate about non-fiction material; influencers credible. 

Profession	Primary purpose	Principal ethical obligation	Independence and conflicts	Relationship with audience
Journalists	Inform the public	Accuracy, verification, fairness, minimizing harm	Must resist pressure from owners, advertisers, political interests and sources	Audience is entitled to expect independent reporting
Commentators	Interpret events and express opinions	Intellectual honesty, factual accuracy, distinguishing evidence from opinion	Should disclose financial, political or institutional interests that could affect their views	Audience should understand the commentator's perspective and possible biases
Advertisers	Promote products, services or causes	Truthful claims, substantiation, avoidance of deceptive or unfair practices	Must disclose sponsorship and avoid misleading representations	Audience is a prospective customer or decision-maker
Marketers	Create demand and influence purchasing decisions	Honest representation, responsible targeting, respect for consumer autonomy and privacy	Must identify material commercial interests and avoid conflicts in recommendations	Audience is a market segment whose needs and behavior are being influenced
Content creators	Produce and distribute information, entertainment or educational material	Originality, attribution, accuracy where factual claims are made, and responsible treatment of subjects	Should disclose sponsorship, conflicts and significant limitations on expertise	Audience may regard the creator as an entertainer, educator, expert or trusted source
Influencers	Shape opinions and behavior through personal credibility and reach	Authenticity, truthful endorsements, clear commercial disclosures	Must disclose payments, gifts, affiliate commissions and other material relationships	Audience often relies on perceived personal experience and trust

In large part, the change in media from physical to virtual has made older issues more difficult. Historically, newspapers and broadcasters generally maintained some separation between their editorial and advertising departments. G

Digital media has weakened that separation. Native advertising, sponsored content, affiliate marketing and influencer endorsements can make commercial persuasion look like independent information.

Monetization issues also make digital content more focused on the ability to attract attention, which arguably diminishes incentives for prudence, honesty, truthfulness or transparency. 

The economic incentives of attention-based media can systematically favor content that is emotionally provocative, divisive, sensational or controversial, as most of us would tend to agree is the case. 

As perhaps always is the case, such ethical guidelines will be honored incompletely, sometimes not at all. Hence the skepticism about the trustworthiness of content whose business model requires sponsorship, advertising or other forms of payment.

That is not to disparage ad support models. As a marketing professor of mine once insisted, "advertising is what makes the free press possible."

But the need for critical thinking remains. People still have to analyze and evaluate claims and information, rather than accepting all or most claims at face value.

AI Infra Winners and Losers from U.S. Midtern Elections

Generally speaking, voters in the 2026 midterm elections seemed to send a message about new AI data centers Late Tuesday evening, 33 ballot measures to restrict data centers were passed by voters or were leading in ongoing vote counts, according to  Data Centers on the Ballot. 

The issue is what that might portend for future AI infrastructure development. Broadly, we might argue that:

• The cost of developing new capacity increases

• Project schedules become less predictable

• The geographic distribution of investment changes

• Costs escalate

• Political risk increases.

The cost of adding new units of compute seemingly is destined to increase, as a result, despite countervailing trends that have been pushing costs down. 

It is too early to argue that the new rules will halt new data center construction. Indeed, most of the rules seem to focus on conditions for such development rather than outright bans. But it might be rational to expect some level of delay in capacity expansion. Projects take longer to approve and complete, but ultimately get built.

Where they get built might well change, though. Companies probably will have to  spend more on power, transmission, cooling, land and community compensation. So areas where terms and conditions are more favorable might get a boost. 

Although all suppliers of AI training and inference operations will face higher obstacles, one might argue that the neocloud suppliers face the most danger. CoreWeave and other neocloud suppliers have less financial flexibility than the hyperscalers; few or no other existing sources of revenue other than available compute capacity and so face cash flow issues if new supply is delayed. 

Company	Construction delays	Higher infrastructure costs	Financing sensitivity	Ability to relocate or adapt
CoreWeave / neoclouds	Very high	High	Very high	Limited to moderate
Amazon	High	High	Moderate	High
Google	High	High	Moderate	High
Oracle	Very high	High	High	Moderate
Microsoft	High	High	Moderate	High
Meta	High	High	Moderate	High

One might also hypothesize that the hyperscalers actually could benefit from the opposition, as smaller developers will face relatively greater hurdles, in the form of:

• More demand for established data center operators with permitted sites

• Greater value for companies with existing power agreements

• More consolidation among neoclouds

• Greater bargaining power for hyperscalers who generate their own power

• Higher barriers to entry for new AI infrastructure competitors.

So among the possible outcomes is some strengthening of the market position of the hyperscalers, at the expense of neocloud providers.

Traditional Media, Social Media and Chatbots All Act as Gatekeepers, But Chatbots are Most User-Driven

As a graduate student in a journalism program I assisted on a study of journalism education. It was pre-internet, so that did not emerge as one of the conclusions. We did not anticipate social media, nor did we imagine chatbots also would emerge. 

But the idea that news media shape the agenda of what people think about is one of the most established concepts in communication research, and so it makes sense that chatbots might emerge in that role as well. 

The central argument is that the media do not necessarily tell people what opinion to hold, but they have considerable influence over which issues people regard as important, which issues receive attention, and which subjects dominate public discussion. Social media has emerged in that role as well.  

AI chatbots may extend it further by synthesizing information and presenting a seemingly-comprehensive explanation of the issues. In other words, it is possible that chatbots shift the terrain from telling people what is happening to influencing what people think is happening and why it matters.

As with seemingly all media these days, the agenda-setting shifts from “what is happening” to something more akin to “what it means,” on an increasingly-personalized basis. 

To be sure, chatbots do not yet have the reach of social or traditional media. A 2025 survey by the Pew Research Center found that 75 percent of respondents say they “never” get news from chatbots. In 2026, Pew finds few Americans asking chatbots for election information. 

  

source: Pew Research

Still, the shift of agenda setting is clear enough:

• Traditional media agenda set by editors and journalists

• Social media agenda set by algorithms

• AI chatbots (an extension of search) set the agenda. 

Traditional news media, social media, and AI chatbots all exercise agenda-setting power: they influence which issues receive attention, which aspects of those issues stand out, and which issues people associate with one another. 

The main difference is how they exercise that power: news organizations select and emphasize stories; social platforms distribute attention through networks and ranking systems; chatbots synthesize information into answers shaped partly by the user’s questions. Of the three platforms, chatbots are the most user-driven. 

Dimension	Traditional news media	Social media	AI chatbots
Main mechanism	Editorial selection, prominence, and repetition	Network circulation, recommendations, and algorithmic ranking	Retrieval, selection, synthesis, and ordering within an answer
Principal gatekeepers	Reporters, editors, owners, and institutional routines	Platform designers, ranking systems, creators, and users	Model developers, training and retrieval systems, behavioral rules, and user prompts
What receives prominence	Stories judged newsworthy	Content made visible through platform and network processes	Topics, explanations, and sources selected for the response
Audience’s role	Choose outlets and respond to coverage	Publish, follow, share, comment, and influence circulation	Ask questions, specify priorities, and revise answers through follow-up prompts
Typical agenda structure	A relatively shared editorial agenda within an outlet	Multiple overlapping, personalized streams	A potentially individualized agenda constructed in conversation

When using traditional news, the audience generally chooses among editorial products: an outlet, program, section, or story. But it is a top-down process, in any case. 

With social media, users help shape their information environment through follows, searches, posts, and interactions, but still encounter streams of material selected and ranked by others. 

With a chatbot, the user can commission a response whose subject and organization need not correspond to any existing article or feed. It is the most bottom-up form of agenda setting.